图像到图像翻译是一类视觉和图形学问题，其目标是通过使用配对的图像对训练集来学习输入图像和输出图像之间的映射。然而，对于许多任务，配对训练数据并不可用。我们提出了一种在缺乏配对示例的情况下学习将图像从源域 X 转换到目标域 Y 的方法。我们的目标是学习一个映射 G: X → Y，使得来自 G(X)的图像分布与 Y 的分布无法区分，这通过对抗损失实现。由于这个映射高度欠约束，我们将其与逆映射 F: Y → X 耦合，并引入循环一致性损失来推动 F(G(X)) ≈ X（反之亦然）。我们在多个不存在配对训练数据的任务上展示了定性结果，包括集合风格迁移、物体变形、季节转换、照片增强等。与几种先前方法的定量比较证明了我们方法的优越性。
Image-to-image translation is a class of vision and graphics problems where the goal is to learn the mapping between an input image and an output image using a training set of aligned image pairs. However, for many tasks, paired training data will not be available. We present an approach for learning to translate an image from a source domain $X$ to a target domain $Y$ in the absence of paired examples. Our goal is to learn a mapping $G: X \rightarrow Y$ such that the distribution of images from $G(X)$ is indistinguishable from the distribution $Y$ using an adversarial loss. Because this mapping is highly under-constrained, we couple it with an inverse mapping $F: Y \rightarrow X$ and introduce a cycle consistency loss to push $F(G(X)) \approx X$ (and vice versa). Qualitative results are presented on several tasks where paired training data does not exist, including collection style transfer, object transfiguration, season transfer, photo enhancement, etc. Quantitative comparisons against several prior methods demonstrate the superiority of our approach.
核心贡献 · Key contributions
提出 CycleGAN，利用循环一致性对抗网络实现无配对图像到图像翻译。 Proposes CycleGAN for unpaired image-to-image translation using cycle-consistent adversarial networks.
引入循环一致性损失，强制前向和后向翻译的一致性。 Introduces cycle consistency loss to enforce forward and backward translation consistency.
在风格迁移、物体变形、季节转换和照片增强等任务上展示有效性。 Demonstrates effectiveness on tasks like style transfer, object transfiguration, season transfer, and photo enhancement.
在定性和定量比较中优于先前的无配对方法。 Outperforms prior unpaired methods in qualitative and quantitative comparisons.